Split the jenkins job up in three parts, so each of them can run in
parallel, and the test part that has to run on a specific node (and
blocks it while running), finishes faster.
Don't install depends of pylint/docs jobs as they will run in docker
and the depends get installed once in the container.
Related: OS#5497
Depends: docker-playground Id5c75725d2fab46b29773fa4f637fa2d73fa7291
Depends: osmo-ci Iea4f15fd9c9f8f36cb8d638c48da000eafe746a4
Change-Id: I5245c529db729e209d78a02ab9c917a90d0e0206
This introduces a hierarchy of classes implementing
* ETS TS 102 225 (general command structure)
* 3GPP TS 31.115 (dialects for SMS-PP)
In this initial patch only the SMS "dialect" is supported,
but it is foreseen that USSD/SMSCB/HTTPS dialects can be
added at a later point.
Change-Id: I193ff4712c8503279c017b4b1324f0c3d38b9f84
With this version I can get all unittests passing:
python -m unittest discover tests/
We're passing argument 'path' to stream_read_entire(), which was
added in [1] and become available since v2.9.51.
Change-Id: I4223c83570d333ad8d79bc2aa2d8bcc580156cff
Related: [1] bfe71315b027e18e62f00ec4de75043992fd2316 construct.git
Related: OS#5666
Argument 'signed' was added in [1] and become available since v2.10.63.
Therefore using bytes2integer() and integer2bytes() from construct.core
bumps the minimum required version of construct to v2.10.63. For
instance, debian:bullseye currently ships v2.10.58.
There is no strict requirement to use construct's API, so let's use
Python's API instead. This allows using older construct versions
from the v2.9.xx family.
Change-Id: I613dbfebe993f9c19003635371941710fc1b1236
Related: [1] 660ddbe2d9a351731ad7976351adbf413809a715 construct.git
Related: OS#5666
-x Print commands and their arguments as they are executed
-e Exit immediately if a command exits with a non-zero status
Change-Id: I13af70ef770936bec00b050b6c4f988e53ee2833
Use multiple processes to speed up pylint. Specifying -j0 will
auto-detect the number of processors available to use.
On AMD Ryzen 7 3700X this significantly reduces the exec time:
$ time python -m pylint -j1 ... pySim *.py
real 0m12.409s
user 0m12.149s
sys 0m0.136s
$ time python -m pylint -j0 ... pySim *.py
real 0m5.541s
user 0m58.496s
sys 0m1.213s
Change-Id: I76d1696c27ddcab358526f807c4a0a7f0d4c85d4
pylint v2.15 is crashing, let's fall-back to a known to work v2.14.5.
Change-Id: Ie29be6ec6631ff2b3d8cd6b2dd9ac0ed8f505e4f
Related: https://github.com/PyCQA/pylint/issues/7375
Related: OS#5668
Register a ProactiveHandler with pySim.transport and call the decoder
from pySim.cat to print a decoded version:
Example usage (exact data only works on my specific card due to the
encrpyted payload):
pySIM-shell (MF/ADF.USIM)> envelope_sms 400881214365877ff6227052000000000302700000201506393535b000118dd46f4ad6b015922f62292350d60af4af191adcbbc35cf4
FETCH: d0378103011300820281838b2c410008812143658700f621027100001c12b000119660ebdb81be189b5e4389e9e7ab2bc0954f963ad869ed7c
SendShortMessage(CommandDetails({'command_number': 1, 'type_of_command': 19, 'command_qualifier': 0}),DeviceIdentities({'source_dev_id': 'uicc', 'dest_dev_id': 'network'}),SMS_TPDU({'tpdu': '410008812143658700f621027100001c12b000119660ebdb81be189b5e4389e9e7ab2bc0954f963ad869ed7c'}))
SW: 9000, data: d0378103011300820281838b2c410008812143658700f621027100001c12b000119660ebdb81be189b5e4389e9e7ab2bc0954f963ad869ed7c
Change-Id: Ia4cdf06a44f46184d0da318bdf67077bc8ac9a1a
This introduces an optional argument to the LinkBase class constructor,
where the application can pass an instance of a ProactiveHandler derived
class in order to handle the proactive commands that the LinkBase is
automatically fetching whenever the card indicates so.
Change-Id: I844504e2fc1b27ce4fc7ede20b2307e698baa0f6
This adds deciding for the bulk of the TLV objects used in the
ETSI CAT (Card Application Toolkit) and 3GPP USAT (USIM Application
Toolkit) systems.
This patch just adds the definitions, but doesn't use them anywhere yet.
Change-Id: I0c66912dbc10164e040e2fec358cef13c45a66ec
The TLV_IE.from_tlv() method is part of a base class that is inherited
by more specific classes. The official way to obtain the tag is the
inherited-class-provided self._compute_tag() method, and *not* a direct
reference to the self.tag member.
This allows for some more obscure TLV parsers, such as the upcoming one
for Proactive Commands in the CAT/OTA context.
Change-Id: I0cd70e31567edc5a0584336efcb5e4282734f6dd
In commit Ib88bb7d12faaac7d149ee1f6379bc128b83bbdd5 I accidentially
broke those commands by adding argparse definitions for better
documentation. When adding the @cmd2.with_argparser decorator,
the method argument changes from the raw string to an argparse.Namespace
object.
This patch fixes the below exception:
pySIM-shell (MF/ADF.USIM)> terminal_profile ffffffff
Traceback (most recent call last):
File "/usr/local/lib/python3.10/dist-packages/cmd2/cmd2.py", line 2129, in onecmd_plus_hooks
stop = self.onecmd(statement, add_to_history=add_to_history)
File "/usr/local/lib/python3.10/dist-packages/cmd2/cmd2.py", line 2559, in onecmd
stop = func(statement)
File "/usr/local/lib/python3.10/dist-packages/cmd2/decorators.py", line 336, in cmd_wrapper
return func(*args_list, **kwargs)
File "/space/home/laforge/projects/git/pysim/pySim/ts_31_102.py", line 1274, in do_terminal_profile
(data, sw) = self._cmd.card._scc.terminal_profile(arg)
File "/space/home/laforge/projects/git/pysim/pySim/commands.py", line 583, in terminal_profile
data_length = len(payload) // 2
TypeError: object of type 'Namespace' has no len()
Change-Id: Ia861eeb2970627d3ecfd0ca73f75ca571c6885b2
Fixes: Ib88bb7d12faaac7d149ee1f6379bc128b83bbdd5
the interpret_sw() method was moved from RuntimeState to RuntimeLchan
in Change-Id I7aa994b625467d4e46a2edd8123240b930305360 - but the code
in pySim/exceptions.py was not adjusted accordingly.
Change-Id: I0614436c99c6a6ebc22c4dc14fb361c5f5f16686
While in the pySim-shell, it's useful to filter the currently selected
file from the choice of available files for select, this doesn't apply
for the tracing case: It's perfectly valid for the UE to SELECT the
file that's already selected right now. The operation basically
becomes equivalent to a STATUS.
Change-Id: I1a20fb3ba70426333ac34448c6cb782c51363965
We need to pass the 'PARENT' flag to get_selectables() to be able
to track SELECT on any of the parent/ancestor DF FID.
Change-Id: Ia7ac627d5edccb97160c90688d720d887fad6ec7
I didn't check the specs, but at least experience with real-world cards
(and modems) shows that it's not just permitted to select the immediate
parent DF, but all ancestors of the currently selected file.
So adjust the get_selectables() method to not just return the immediate
parent, but to recurse all the way up and report the FID of any ancestor
DF.
Change-Id: Ic9037aa9a13af6fb0c2c22b673aa4afa78575b49
In order to be able to explicitly select the MF via 3f00,
we need to pass the 'MF' to get_selectables(), so the record
is included in the list of selectable files from the current
working directory.
Change-Id: I27085896142fe547a6e93e01e63e59bbc65c8b8a
All other get_selectables() understand a flag like 'FIDS' to request
only the hexadecimal FIDs and not the file names. However, the
CardEF.get_selectables() ignored those flags and unconditionally
returned the names.
Change-Id: Icdc37cae3eecd36d167da76c30224b9d48c844fd
The Tracer implemented those options and the argparser handled it,
but we didn't ever connect the two.
Change-Id: I7d7d5fc475a8d09efdb63d3d6f1cc1de1996687b
If the modem/UE doesn't ask for the FCP to be returned, a SELECT
can exit with 9000 and no response body. Don't crash in that case.
Change-Id: I66788717bec921bc54575e60f3f81adc80584dbc
Instead of a cryptic backtrace, we now return a meaningful error like this:
{"error": {"message": "Security Status not satisfied - Card PIN enabled?", "status_word": "6982"}
Change-Id: I6dafd37dfd9fa3d52ca2c2e5ec37a6d274ba651b
Closes: OS#5606
Let's make sure even error messages are returned in JSON format.
While at it, also reduce some code duplication between the 'auth'
and 'info' route handlers by using the klein handle_errors decorator
instead of manual exception catching.
Change-Id: I1e0364e28ba7ce7451993f57c8228f9a7ade6b0e
Closes: OS#5607
This introduces a new pySim.apdu module hierarchy, which contains
classes that represent TPDU/APDUs as exchanged between
SIM/UICC/USIM/ISIM card and UE.
It contains instruction level decoders for SELECT, READ BINARY and
friends, and then uses the pySim.filesystem.Runtime{Lchan,State} classes
to keep track of the currently selected EF/DF/ADF for each logical
channel, and uses the file-specific decoder classes of pySim to decode
the actual file content that is being read or written.
This provides a much more meaningful decode of protocol traces than
wireshark will ever be able to give us.
Furthermore, there's the new pySim.apdu_source set of classes which
provides "input plugins" for obtaining APDU traces in a variety of
formats. So far, GSMTAP UDP live capture and pyshark based RSPRO
live and pcap file reading are imlpemented.
Change-Id: I862d93163d495a294364168f7818641e47b18c0a
Closes: OS#5126
Fix a bug in the pySim.sysmocom_sja2 module, where we defined unnamed
bits in BitStruct without a default value causing exceptions like this:
EXCEPTION of type 'KeyError' occurred with message: 'None'
Change-Id: Ib2da5adda4fae374ab14bb8100f338691aef719a
Closes: OS#5575
We were invoking the constructor with the description as 4th positional
argument, but that was actually the 'size' argument in this case.
Let's swap the order to be aligned with other file constructors.
Change-Id: I9acee757f096fef0d8bacbec3b52f56267cd52f6
The size should be a *tuple*. In reality we so far passed a set. The
problem with the set is that ordering is not guaranteed, and hence we
cannot assume the first and second item have meaning (minimum vs.
default record length).
Change-Id: I470f4e69c83cb2761861b3350bf8d49e31f4d957
As the documentation strings say: The size should be a *tuple*. In
reality we so far passed a set. The problem with the set is that
ordering is not guaranteed, and hence we cannot assume the first and
second item have meaning (minimum vs. default size).
While at it, use a type annotation to catch such bugs easily.
Change-Id: I553616f8c6c4aaa8f635b3d7d94e8e8f49ed5a56
Reduce all the copy+pasted '/'.join(path_list) constructs with
a method returning the formatted path string.
Change-Id: I5e9bfb425c3a3fade13ca4ccd2b891a0c21ed56d
Otherwise we have binary/bytes as values inside the dict, rather than a
hexadecimal string. That's ugly when printing without json formatting.
Change-Id: Ia3e7c4791d11bd4e3719a43d58e11e05ec986d1f
Even while we don't yet have a proper decoder, let's at least represent
the network name as hex-string
Change-Id: I4ed626699d1e4e484d4ffd04349676dadff626a0
This file existed in earlier specs like Release 3.8.0, but was removed
in later revisions. Still, there are cards around implementing that
older spec, so let's add a decoder.
Change-Id: Ic7163b2a01f64ef1223cf15b8d0813d3edf5b61a
Christian Amsüss
